[PATCH, ARM] PR68674 Fix LTO support for neon builtin and error catching (ping)

[Christian Bruel]

It seems that changing the arm_cpu_builtins code in arm-c.c to do:
     cpp_undef (pfile, "__ARM_NEON_FP");
     if (TARGET_NEON_FP)
       builtin_define_with_int_value ("__ARM_NEON_FP", TARGET_NEON_FP);

instead of:
     if (TARGET_NEON_FP)
       builtin_define_with_int_value ("__ARM_NEON_FP", TARGET_NEON_FP);
     else
       cpp_undef (pfile, "__ARM_NEON_FP");

fixes this. I guess we should be undefing any macros before redefining them.I

>> I don't know, If we are going from v7 to v8, the __ARM_NEON_FP value indeed changes. The question is whether we want to hide this to the user ?
> Well, the user explicitly changed the fpu level with the pragma in the testcase, which rightly has the effect of
> changing the value of some predefines. I don't think warning is appropriate here, as long as the predefine
> has the right value in the right pragma scope.
>
note that the issue can be reproduced on the current trunc with the 
default configure with

arm-none-eabi-gcc -mfloat-abi=softfp  -mfpu=neon-fp-armv8
--------------------------------------------------
#pragma GCC target ("fpu=neon")
-------------------------------------------------
that gives:

warning: "__ARM_NEON_FP" redefined
  <built-in>: note: this is the location of the previous definition

So I agree, the warning seems intuitively useless, but I'd just need to 
mention :

iso/iec 99 A.3 and 6.10 (Preprocessing directives) #pragma is a 
preprocessor directive
and iso/iec 99 6.10.3 (Macro Replacement)

also, you fix is slightly wrong, since the warning should be kept for 
other explicit #define user redefinitions. And I think we shouldn't use 
cpp_undef for that. It's better to use NODE_CONDITIONAL flags on the 
macro identifier, since this is what it is.

I'd like to send a patch for this separately, so you can continue 
reviewing this one independently of the decision

note that for the time being to make the patches independent the current 
test can be fixed with

#if __ARM_ARCH==8
#pragma GCC target ("fpu=neon-fp-armv8")
#else
#pragma GCC target ("fpu=neon")
#endif

thanks for catching this with the aarch32 v8 validation. I'll add this 
option to my validation scripts.

Cheers